The Cognitive Learning for Vision and Robotics (CLVR) Lab, led by Joseph J. Lim at the Korea Advanced Institute of Science and Technology (KAIST) in Seoul, focuses on developing intelligent systems. These systems make sequential decisions through perception, action, and reasoning. Research areas include reinforcement learning, robot learning, world model with representation learning, visual perception, and symbolic manipulation.

한국과학기술원(KAIST) 임재환 교수가 이끄는 인지 비전 및 로봇 공학(CLVR) 연구실은 지능형 시스템 개발에 중점을 둔다. 해당 시스템은 인식, 행동 및 추론을 통해 순차적 의사 결정을 수행하며, 강화 학습, 로봇 학습, 표현 학습 기반의 월드 모델, 시각적 인식 및 상징적 조작 등의 연구 분야를 포함한다.

ReWiND: Language-Guided Rewards Teach Robot Policies without New Demonstrations

Mitigating Suboptimality of Deterministic Policy Gradients in Complex Q-functions

Subtask-Aware Visual Reward Learning from Segmented Demonstrations

QMP: Q-switch Mixture of Policies for Multi-Task Behavior Sharing

EXTRACT: Efficient Policy Learning by Extracting Transferrable Robot Skills from Offline Data

DROID: A Large-Scale In-the-Wild Robot Manipulation Dataset

Open X-Embodiment: Robotic Learning Datasets and RT-X Models

SPRINT: Scalable Policy Pre-Training via Language Instruction Relabeling

Bootstrap Your Own Skills: Learning to Solve New Tasks with Large Language Model Guidance

FurnitureBench: Reproducible Real-World Benchmark for Long-Horizon Complex Manipulation

PATO: Policy Assisted TeleOperation for Scalable Robot Data Collection

Hierarchical Neural Program Synthesis

Cross-Domain Transfer via Semantic Skill Imitation

Skill-based Model-based Reinforcement Learning

Skill-based Meta-Reinforcement Learning

Task-Induced Representation Learning

Know Your Action Set: Learning Action Relations for Reinforcement Learning

Learning to Synthesize Programs as Interpretable and Generalizable Policies

Generalizable Imitation Learning from Observation via Inferring Goal Proximity

Demonstration-Guided Reinforcement Learning with Learned Skills

Adversarial Skill Chaining for Long-Horizon Robot Manipulation via Terminal State Regularization

Distilling Motion-Planner Augmented Policies into Visual Control Policies for Robot Manipulation

Policy Transfer across Visual and Dynamics Domain Gaps via Iterative Grounding

IKEA Furniture Assembly Environment for Long-Horizon Complex Manipulation Tasks

Message Passing Adaptive Resonance Theory for Online Active Semi-supervised Learning

Accelerating Reinforcement Learning with Learned Skill Priors

Motion Planner Augmented Reinforcement Learning for Robot Manipulation in Obstructed Environments

Generalization to New Actions in Reinforcement Learning

Keyframing the Future: Keyframe Discovery for Visual Prediction and Planning

Program Guided Agent

Learning to Coordinate Manipulation Skills via Skill Behavior Diversification

To Follow or not to Follow: Selective Imitation Learning from Observations

Multimodal Model-Agnostic Meta-Learning via Task-Aware Modulation

Composing Complex Skills by Learning Transition Policies

Scaling Simulation-to-Real Transfer by Learning Composable Robot Skills

Multi-view to Novel view: Synthesizing Novel Views with Self-Learned Confidence

Neural Program Synthesis from Diverse Demonstration Videos

Demo2Vec: Reasoning Object Affordances From Online Videos

Learning and Using the Arrow of Time

Universal Agent for Disentangling Environments and Tasks

Multi-Modal Imitation Learning from Unstructured Demonstrations using Generative Adversarial Nets

Unsupervised Visual-Linguistic Reference Resolution in Instructional Videos

Knowledge Acquisition for Visual Question Answering via Iterative Querying

Target-driven Visual Navigation in Indoor Scenes using Deep Reinforcement Learning

High-Fidelity Facial and Speech Animation for VR HMDs

Single Image 3D Interpreter Network

Physics 101: Learning Physical Object Properties from Unlabeled Videos

Galileo: Perceiving Physical Object Properties by Integrating a Physics Engine with Deep Learning

Discovering States and Transformations in Image Collections

FPM: Fine pose Parts-based Model with 3D CAD models

Looking Beyond the Visible Scene

Parsing IKEA Objects: Fine Pose Estimation

Sketch Tokens: A Learned Mid-level Representation for Contour and Object Detection

Transfer Learning by Borrowing Examples for Multiclass Object Detection

Experimental Evaluation of Support Vector Machine-based and Correlation-based Approaches to Automatic Particle Selection

Exploiting Hierarchical Context on a Large Database of Object Categories

Context by Region Ancestry

Recognition using Regions

A Distributed Message Passing Algorithm for Sensor Localization

Target-driven visual navigation in indoor scenes using deep reinforcement learning

Open X-Embodiment: Robotic Learning Datasets and RT-X Models : Open X-Embodiment Collaboration 0

Recognition using regions

Sketch tokens: A learned mid-level representation for contour and object detection

Galileo: Perceiving physical object properties by integrating a physics engine with deep learning

Exploiting hierarchical context on a large database of object categories

Droid: A large-scale in-the-wild robot manipulation dataset

Learning and using the arrow of time

Open x-embodiment: Robotic learning datasets and rt-x models

Discovering states and transformations in image collections

Single image 3d interpreter network

Accelerating reinforcement learning with learned skill priors

Multimodal model-agnostic meta-learning via task-aware modulation

Multi-modal imitation learning from unstructured demonstrations using generative adversarial nets

Cognitive Learning for Vision and Robotics (CLVR) lab

교수님 소개

임재환

연구실 소개

연구실 키워드

알럼나이가 간 회사

연구실 지원 방법

대우 조건

연도별 논문 인용수

저널 & 학회 논문 출판 수

연구실 논문 목록

ReWiND: Language-Guided Rewards Teach Robot Policies without New Demonstrations

Mitigating Suboptimality of Deterministic Policy Gradients in Complex Q-functions

Subtask-Aware Visual Reward Learning from Segmented Demonstrations

연구실 보유 특허

대학별 입시끈

우리 연구실 모집

교수님 소개

임재환

연구실 키워드

알럼나이가 간 회사

연구실 지원 방법

대우 조건

연도별 논문 인용수

저널 & 학회 논문 출판 수

연구실 논문 목록

ReWiND: Language-Guided Rewards Teach Robot Policies without New Demonstrations

Mitigating Suboptimality of Deterministic Policy Gradients in Complex Q-functions

Subtask-Aware Visual Reward Learning from Segmented Demonstrations

연구실 보유 특허

최근 3년간 연구실적 높은 연구실

대학별 입시끈

우리 연구실 모집